NVIDIA Tribute to David Blackwell (1919 - 2010) (Re: Single-paradigm Logic Programming: CogitoAI System)

From	Mild Shock <janburse@fastmail.fm>
Newsgroups	sci.logic
Subject	NVIDIA Tribute to David Blackwell (1919 - 2010) (Re: Single-paradigm Logic Programming: CogitoAI System)
Date	2026-05-07 12:02 +0200
Message-ID	<10thnuq$9h5n$1@solani.org> (permalink)
References	<10arigv$oe7h$4@solani.org> <10rqkp2$dal8$2@solani.org>

Show all headers | View raw

Hi,

Now I get finger licking feelings to get my
hands on some of NVIDIA DGX or something
machines with 1000 TOPS or something.

Interestingly their Blackwell 2024 series,
is a tribute to David Blackwell (1919 - 2010),
who co-authored typical post WWII and

pre Cold War "strategic" mathematics:

Bayes and Minimax Solutions of Sequential Decision Problems
Author(s): K. J. Arrow, D. Blackwell and M. A. Girshick
Source: Econometrica, Vol. 17, No. 3/4 (Jul. - Oct., 1949), pp. 213-244

Is this basic statistics or neuro symbolic?

Bye

P.S.: Nice and maybe easy special case:

3. OPTIMUM SEQUENTIAL PROCEDURE FOR A
DICHOTOMY WHEN THE COST FUNCTION IS LINEAR
We are given two alternative hypotheses H1 and H2,
which, for the  sake of simplicity, we assume are
characterized respectively by two probability densities
f1(x) and f2(x) of a random vector X in an R dimensional
Euclidean space. Etc.. Etc..

Mild Shock schrieb:
> Hi,
> 
> The Latin cogito, ergo sum, usually translated
> into English as "I think, therefore I am",[a]
> is the "first principle" of the philosophy of
> the French scientist and philosopher René Descartes.
> 
> The new CogitoAI System features explanation
> compression using the motto:
> 
> **********************************************
> Search First, Explain Later
> **********************************************
> 
> Its very easy to implement, and based on intuitionist
> logic Curry Howard exists witness terms, a rule:
> 
> P(X, Y) :- Q(X, Z), R(Z, Y).
> 
> Is transformed into, combining existential and conjunction:
> 
> P(X, Y, rule(Z, S, T)) :- Q(X, Z, S), R(Z, Y, T).
> 
> So the search will only return the compressed form
> you can later generate fancy trees, with fact
> prevenance, rule references, coloring, bells and
> 
> whistles. Dunno, just an idea based on:
> 
> formalise existential quantification by
> declaring a suitable record type
> https://plfa.github.io/Quantifiers/#existentials
> 
> LoL
> 
> Bye
> 
> Mild Shock schrieb:
>> Hi,
>>
>> The diagram figure 2 here:
>>
>> Fifty Years of Prolog and Beyond (TPLP 2022)
>> KÖRNER P, LEUSCHEL M, BARBOSA J, et al. Fifty Years of Prolog and 
>> Beyond. Theory and Practice of Logic Programming.
>> 2022;22(6):776-858. doi:10.1017/S1471068422000102
>>
>> And reproduced here:
>>
>> Comparison of Prolog implementations
>> The page has also missing Scryer Prolog, which
>> I think is an important Prolog system written in
>> Rust, because it also pays tribute to Prolog II.
>> https://en.wikipedia.org/wiki/Comparison_of_Prolog_implementations
>>
>> Is pretty much brainwashed nonsense. Most Prolog
>> systems, that have cyclic terms, are derived from
>> Prolog II.  Adopting a non-canonical rational tree
>>
>> term approach. This includes:
>>
>> - SICStus Prolog
>> - Ciao Prolog
>> - YAP Prolog
>> - SWI-Prolog
>> - Scryer Prolog
>> - Trealla Prolog
>> - Dogelog Player
>> - What else?
>>
>> SICStus Prolog is possibly the most advanced, (*)
>> it also supports asserts and copying, whereas I found
>> not all Prolog systems listed above can even
>>
>> copy cyclic terms, despite they can unify them.
>> Basically the philogeny of Prolog systems is
>> not some "single inheritance" tree. Cyclic terms
>>
>> algorithm have nice side effect that they might
>> speed up acyclic term arguments as well. But
>> cyclic terms is one of the topics that is very
>>
>> badily  covered, for some individuals difficult (**)
>> to understand and sometimes even completely ignored.
>>
>> Bye
>>
>> (*)
>> SICStus Prolog unifies, compares (see ref-lte-cte),
>> asserts, and copies cyclic terms without looping.
>> The write_term/[2,3] built-in predicate can
>> optionally handle cyclic terms.
>> https://sicstus.sics.se/sicstus/docs/4.6.0/html/sicstus/ref_002dsem_002docc.html 
>>
>>
>> (**)
>> Because of the infinite looping, their brains might
>> also get into infinite loops. Even fuzzy testing does
>> not help anymore breaking these loops.
>

Back to sci.logic | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

Single-paradigm Logic Programming: CogitoAI System (Was: Cyclic terms are missing from "Fifty Years of Prolog and Beyond (TPLP 2022)") Mild Shock <janburse@fastmail.fm> - 2026-04-16 14:30 +0200
  DeepSeek nearly as stupid as ChatGPT (Re: Single-paradigm Logic Programming: CogitoAI System) Mild Shock <janburse@fastmail.fm> - 2026-04-16 14:44 +0200
    Zero Storage for Witnesses: No Bread Crumbs (Re: DeepSeek nearly as stupid as ChatGPT) Mild Shock <janburse@fastmail.fm> - 2026-04-16 15:00 +0200
  NVIDIA Tribute to David Blackwell (1919 - 2010) (Re: Single-paradigm Logic Programming: CogitoAI System) Mild Shock <janburse@fastmail.fm> - 2026-05-07 12:02 +0200
    Hunting the Hanta Virus (MV Hondius) (Re: NVIDIA Tribute to David Blackwell (1919 - 2010)) Mild Shock <janburse@fastmail.fm> - 2026-05-07 12:49 +0200

csiph-web